Clinical and pathological studies indicate that the hindbrain, particularly the brainstem, has a greater tolerance to ischemia than the forebrain structures. In patients with basilar artery thrombosis, it is this hypothesis that provides the rationale for the current practice of revascularization up to 24 hours after symptom onset. If there is a benefit for these patients, whatever the mechanism (relative ischemic tolerance, collateral circulation), it is very likely that patients with thrombus in the branches of the basilar artery would benefit as well. Currently, the benefit of such delayed reperfusion with tPA is unproven in either population. This group represents a large number of patients: 10 to 20% of all ischemic strokes are located in the brainstem or cerebellum. The study is a pilot, single-center, randomized, prospective, blinded safety and feasibility trial of 0.9 mg/kg iv t-PA given over one hour between 3 to 24 hours after onset of brainstem and/or cerebellar ischemic stroke. After a computed tomographic (CT) scan to exclude hemorrhage, a magnetic resonance (MR) study will be obtained to confirm brainstem and/or cerebellar stroke. Sixty patients will be randomized evenly to treatment or control arms. The primary safety endpoint will be symptomatic intracranial hemorrhage within 36 hours after infusion. CT scans will be obtained if neurological decline occurs. Secondary aims of the present study are to obtain preliminary data on efficacy (Aim 2). Twenty-four hour post infusion, 7-10 day, and 90-day functional outcome will be recorded. The rate of improvement by >8 points on the NIH Stroke Scale at 24 hours and functional outcome at 90 days will be compared between treatment groups. Repeat MR scans will be obtained 24 hours (vessel recanalization, DWI lesion growth) and 90 days (final infarct volume) after tPA infusion in all patients. The rate growth of the DWI lesion volume on 24 hour MR, vessel recanalization by MRA at 24 hours, the final lesion volume and lesion growth at 90 day MR will be compared between groups. Other secondary aims are to investigate clinical and imaging predictors of outcome (Aim 3), and to assess the accuracy of the Oxfordshire criteria (Aim 4) for the identification of patients with brainstem and cerebellar stroke. These data will be used to design and plan further investigation in this patient population.